Copy-up (OverlayFS)
Copy-up creates an upper-layer counterpart for a lower-layer object when an OverlayFS operation requires it.
Writable opens, metadata changes and hard-link creation can trigger copy-up. Metadata-only copy-up can defer regular-file data, so an upper entry is not proof that its payload has moved. [1]
Linux v6.12 tries cloning before copying file data. A new upper file therefore need not have independently allocated data blocks. A truncating open can set the transferred length to zero; copy-up is not an archival preservation operation. [2]
Relations
- related
- OverlayFS, Metadata-only copy-up
Advantages
- Allows changes through the upper layer without rewriting the lower object.
Limitations
- The operation and configured features determine which metadata and data move.
- An upper-layer counterpart is not by itself a complete backup.
Citations
- Linux 6.12: Overlay Filesystem, non-directories and metadata-only copy-up
- Linux v6.12: OverlayFS copy-up eligibility, data transfer and truncation
Sources & caveats
- Source review of Linux 6.12 documentation and implementation; no mounted-overlay, restore or crash experiment.